How to determine relationship between 2 or more variables
A better understanding of correlation is required to better interpret relationship between variables
Scatterplots are great to display relationships between 2 variables. However, how can you show there is indeed a meaningful relationship between the 2 variables?
If you want to test if the 2 variables follows a linear relationship, Pearson and Spearman correlation can be used. If there are no outliers, Pearson correlation can be used. Otherwise, choose Spearman correlation instead.
The correlation coefficient value indicate extent of linearity, where value of 1 will indicate perfect positive correlation and -1 will indicate perfect negative correlation. The p-value indicates if the correlation is significant. To show if there is linear correlation between multiple variables, correlation matrices or pair plots can be plotted.
As an illustration, we used correlation matrix to show that the self-amplifying mRNA vaccine (ARCT-021) displays innate transcriptomic responses that are correlated with the other licensed vaccines. The size and intensity of the colour shows the correlation coefficient value (see figure above). This correlation matrix was plotted in R, using the cor() and corrplot() libraries. For more…
